Thursday, December 23, 2010

MultiStage (v7) System Results - Part Eight (Final)

Over the past several weeks I've posted a series of articles describing the results of my current trading system. I provided this detailed look for a few reasons:
  • To describe what results are interesting to me: If you are a system developer, or thinking about system development, sometimes it's useful to look at things from another person's perspective. What do I look at? Why is it important?
  • To show what results can look like: When I was first developing trading systems I often had results that I didn't know how to assess. Is averaging 1% per trade good? Is a Sharpe Ratio of 1.2 good? Having something to compare against can often be very helpful because it provides a base-line.
  • To describe to Covestor.com followers what I do: I'm not sure if you have followed any of my trades or model at Covestor.com, but you should. It's a wonderful service where people can actually mirror the trades of an investor. This series allowed me to show some of my backtesting and expectations to those who are mirroring (or thinking about mirroring) my account.
In the series I tried to provide a decent overview of what I experience in backtesting. Naturally, I always provide the caveat that backtesting is no guarantee that we will experience similar results in the future. It only shows what has happened in the past. That being said, most of the testing I showed was quite extensive. Generally I show tests of a decade or more, consisting of thousands of trades.

In Part One I provided an overview of the key statistics from the MultiStage v7 Trading System. By the way, v7 refers to the current version. This current system has undergone continual improvements and is always a work in progress.

Part Two was a month-by-month look at the results. This is something that is helpful to look at because it gives some idea of consistency. It is supplemented by Part Three and Part Four(showing quarterly results), which again, stress and demonstrate consistency of past results.

Part Five is all about draw-downs. I attempted to explain what draw-downs were, why they were important, and how I measure them. Finally, I showed several statistics describing the draw-downs of the MultiStage v7 Trading System.

Part Six and Part Seven are a summary of additional performance statistics that are not CAGR (the stat everyone loves to look at). Rather they are important data that describe much more of the character of the trading system. How often does it trade? What does volatility look like? How much does a typical trade earn? How long is it held? And more.

Finally, I posted an addendum about how the system would have performed in the decade prior to 2000. This was sparked by a post on MarketSci, which showed that short term mean reversion systems, while they have worked since 2000, did not do well in the prior decade. I was curious about how my system would have performed, since it has some mean reversion components.

A Note About Parameters:

While all of these results come from a single trading system, there are a number of parameters or inputs that can be adjusted when running backtests. For example, if the trade calls for a stop to be in place I might adjust it to 3%, 4%, 5%, or some other variable based on volatility. These parameters tweak the results a little, but the overall character of the system is still the same.

I hope this summary of results is useful for the reasons mentioned above. If there are other statistics you are interested in please let me know.

Good Trading...

No comments:

Post a Comment